Abstract: The present invention relates to generating training data for computer models, specifically video detections models, using transfer learning techniques. More particularly, the present invention relates to downsampling and upscaling substantially high-quality training data for the training of a new computer model to be used as a video detection model. Aspects and/or embodiments seek to provide a system and/or method of using transfer learning techniques to generate training data for new computer models in security event detection methods and/or systems using image data from surveillance cameras as input data.